李新春 Xin-Chun Li Ph.D. student LAMDA Group School of Artificial Intelligence Nanjing University, Nanjing 210023, China. Email: lixc@lamda.nju.edu.cn

Short Biography

I received my B.Sc. degree in School of Information Management from Nanjing University, China in June 2018.

I received my M.Sc. degree in Department of Computer Science and Technology from Nanjing University, China in June 2021.

In the same year, I was admitted to study for a Ph.D. degree in School of Artificial Intelligence from Nanjing University, which is also in the LAMDA Group led by professor Zhi-Hua Zhou, under the supervision of Prof. De-Chuan Zhan and Prof. Ming Li.

For more information, refer to my CH_CV, Google Scholar, DBLP.

Main Research Interests

• Landscape Analysis and Visualization

• Model Fusion, Monotonic Linear Interpolation, Mode Connectivity

• Distributed Model Reuse

• Learnware Specification

• Federated Learning

• Ensemble Learning

• Transfer Learning, Domain Adaptation

• Meta Learning, Few-shot Learning

• Reinforcement Learning

For more information, refer to my Zhihu and Github.

Manuscripts

• Xin-Chun Li , Wen-Shu Fan, Bowen Tao, Le Gan, De-Chuan Zhan. Exploring Dark Knowledge under Various Teacher Capacities and Addressing Capacity Mismatch. Arxiv 2405.13078. [paper]

• Xin-Chun Li , Lan Li, De-Chuan Zhan. Visualizing, Rethinking, and Mining the Loss Landscape of Deep Neural Networks. Arxiv 2405.12493. [paper]

Technical Reports

• Xin-Chun Li , Le Gan, De-Chuan Zhan, Yunfeng Shao, Bingshuai Li, Shaoming Song. Aggregate or Not? Exploring Where to Privatize in DNN Based Federated Learning Under Different Non-IID Scenes. CoRR 2021. [paper] [codes]

• Xin-Chun Li , Lan Li, De-Chuan Zhan, Yunfeng Shao, Bingshuai Li, Shaoming Song. Preliminary Steps Towards Federated Sentiment Classification. CoRR 2021. [paper]

Conference Publications

• Xin-Chun Li , Jin-Lin Tang, Bo Zhang, Lan Li, De-Chuan Zhan. Exploring and Exploiting the Asymmetric Valley of Deep Neural Networks. In: Advances in Neural Information Processing Systems 37 (NeurIPS'24), Vancouver, Canada, 2024. [paper]

• Yichu Xu, Xin-Chun Li, Le Gan, De-Chuan Zhan. Weight Scope Alignment: A Frustratingly Easy Method for Model Merging. In: Proceedings of the 27TH EUROPEAN CONFERENCE ON ARTIFICIAL INTELLIGENCE (ECAI'24), Santiago de Compostela, 2024. [paper] (CCF-B会议, 第二作者)

• Bowen Tao^*, Xin-Chun Li^* , De-Chuan Zhan. MLI Formula: A Nearly Scale-Invariant Solution with Noise Perturbation. In: Proceedings of the 41st International Conference on Machine Learning (ICML'24), Vienna, Austria, 2024. [paper] (CCF-A会议, 共同一作)

• Lan Li, Xin-Chun Li , Han-Jia Ye, De-Chuan Zhan. Enhancing Class-Imbalanced Learning with Pre-trained Guidance through Class-Conditional Knowledge Distillation. In: Proceedings of the 41st International Conference on Machine Learning (ICML'24), Vienna, Austria, 2024. [paper] (CCF-A会议, 第二作者)

• Wen-Shu Fan, Su Lu, Xin-Chun Li , De-Chuan Zhan, Le Gan. Revisit the Essence of Distilling Knowledge through Calibration. In: Proceedings of the 41st International Conference on Machine Learning (ICML'24), Vienna, Austria, 2024. [paper] (CCF-A会议, 第三作者)

• Bowen Tao, Lan Li, Xin-Chun Li, De-Chuan Zhan. CLAF: Contrastive Learning with Augmented Features for Imbalanced Semi-Supervised Learning. In: Proceedings of the 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP'24), Seoul, Korea, 2024. [paper] (CCF-B会议, 第三作者)

• Xin-Chun Li , Wen-Shu Fan, Shaoming Song, Yinchuan Li, Bingshuai Li, Yunfeng Shao, De-Chuan Zhan. Asymmetric Temperature Scaling Makes Larger Networks Teach Well Again. In: Advances in Neural Information Processing Systems 35 (NeurIPS'22). [paper] [codes] (CCF-A会议, 第一作者)

• Xin-Chun Li , Jin-Lin Tang, Shaoming Song, Bingshuai Li, Yinchuan Li, Yunfeng Shao, Le Gan, De-Chuan Zhan. Avoid Overfitting User Specific Information in Federated Keyword Spotting. In: Proceedings of the 23rd INTERSPEECH Conference (INTERSPEECH'22), online conference, Songdo ConvensiA, Incheon, Korea, 2022. [paper] [codes] (CCF-C会议, 第一作者)

• Xin-Chun Li , Yi-Chu Xu, Shaoming Song, Bingshuai Li, Yinchuan Li, Yunfeng Shao, De-Chuan Zhan. Federated Learning with Position-Aware Neurons. In: Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR'22), online conference, New Orleans, Louisiana, 2022. [paper] [codes] (CCF-A会议, 第一作者)

• Xin-Chun Li , Yan-Jia Wang, Le Gan, De-Chuan Zhan. Exploring Transferability Measures and Domain Selection in Cross-Domain Slot Filling. In: Proceedings of the 2022 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP'22), online conference, Singapore, 2022. [paper] [codes] (CCF-B会议, 第一作者)

• Xin-Chun Li , De-Chuan Zhan, Yunfeng Shao, Bingshuai Li, Shaoming Song. FedPHP: Federated Personalization with Inherited Private Models. In: Proceedings of the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD'21), online conference, Bilbao, Spain, 2021. [paper] [codes] (CCF-B会议, 第一作者)

• Xin-Chun Li , De-Chuan Zhan. FedRS: Federated Learning with Restricted Softmax for Label Distribution Non-IID Data. In: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD'21), online conference, Singapore, 2021. [paper] [codes] (CCF-A会议, 第一作者)

• Han-Jia Ye, Xin-Chun Li , De-Chuan Zhan. Task Cooperation for Semi-Supervised Few-Shot Learning. In: Proceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI'21), online conference, 2021. [paper] [codes] (CCF-A会议, 第二作者, 学生一作)

• Xin-Chun Li , De-Chuan Zhan, Jia-Qi Yang, Yi Shi, Cheng Hang, Yi Lu. Towards Understanding Transfer Learning Algorithms Using Meta Transfer Features. In: Proceedings of the 24th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD'20), online conference, Singapore, 2020. [paper] (CCF-C会议, 第一作者)

• Jia-Qi Yang, De-Chuan Zhan, Xin-Chun Li . Bottom-Up and Top-Down Graph Pooling. In: Proceedings of the 24th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD'20), online conference, Singapore, 2020. [paper] (CCF-C会议, 第三作者, 学生二作)

• Zhao-Yang Fu, De-Chuan Zhan, Xin-Chun Li , Yi-Xing Lu. Automatic Successive Reinforcement Learning with Multiple Auxiliary Rewards. In: Proceedings of the 28th International Joint Conference on Artificial Intelligence (IJCAI'19), Macao, China, 2019. [paper] (CCF-A会议, 第三作者, 学生二作)

Journal Publications

• Xin-Chun Li , Shaoming Song, Yinchuan Li, Bingshuai Li, Yunfeng Shao, Yang Yang, De-Chuan Zhan. MAP: Model Aggregation and Personalization in Federated Learning With Incomplete Classes. In: IEEE Transactions on Knowledge and Data Engineering. DOI: 10.1109/TKDE.2024.3390041. [paper] [codes] (CCF-A期刊, SCI, 第一作者)

• Xin-Chun Li , Yang Yang, De-Chuan Zhan. MrTF: Model Refinery for Transductive Federated Learning. In: Data Mining and Knowledge Discovery. 2023, 37(5):2046-2069. [paper] (CCF-B期刊, SCI, 第一作者)

• Lan Li, De-Chuan Zhan, Xin-Chun Li . Aligning Model Outputs for Class Imbalanced Non-IID Federated Learning. In: Machine Learning. 2022:1-24. [paper] (CCF-B期刊, SCI, 第三作者, 学生二作)

• Xin-Chun Li , De-Chuan Zhan. Distributed Model Reuse with Multiple Classifier. CCML'2021. [paper] (国内会议, 第一作者)

  李新春 , 詹德川. 使用多分类器的分布式模型重用技术. 计算机科学与探索, 2021. (国内CCF-B期刊, 第一作者)

• Xin-Chun Li , De-Chuan Zhan, Jia-Qi Yang, Yi Shi. Deep Multiple Instance Selection. In: Science China Information Sciences. Vol. 64 No. 3. [paper] [codes] (CCF-A期刊, SCI, 第一作者)

• Xin-Chun Li , De-Chuan Zhan. A semantic relation preserved word embedding reuse method. Scientia Sinica: Informationis. 2020, 50(6):813-823. [paper] (国内CCF-A期刊, 第一作者)

  李新春 , 詹德川. 一种保持语义关系的词向量复用方法. 中国科学：信息科学, 2020, 50(6):813-823.

Publications Summary

First Author	CCF-A	CCF-B	CCF-C	Others
Conference	KDD'21, CVPR'22, NeurIPS'22, ICML'24, NeurIPS'24	ECML-PKDD'21, ICASSP'22	PAKDD'20, INTERSPEECH'22	-
Journal	SCIS'21, TKDE'24	DMKD'23	-	《中国科学：信息科学》'20, 《计算机科学与探索》'21
Not First Author	CCF-A	CCF-B	CCF-C	Others
Conference	IJCAI'19, AAAI'21, ICML'24, ICML'24	ECAI'24, ICASSP'24	PAKDD'20	-
Journal	-	Machine Learning	-	-

Code Repositories

• [FedRepo] [FaMnist] [CIFAR10] [CIFAR100]

• [ATS-LargeKD] [CIFAR100] [RES14] [RES110] [WRN28-1] [WRN28-8]

Invited Presentations

• 中国电科14所技术交流. 人工智能技术及其应用. 2022.08.17. [pdf]
• 南京智谷人工智能研究院. 人工智能技术及其应用. 2022.06.10. [pdf]
• 第四范式技术交流. NonIID数据下的联邦模型聚合和模型个性化. 2021.12.03. [pdf]
• 华为诺亚联邦学习Workshop. 基于私有共享模型的联邦学习个性化技术. 2021.07.30. [pdf]
• 华为NAIE技术大讲堂第31期. 基于相似性度量的可迁移性分析和迁移学习方法. 2021.03.18. [pdf]
• 华为云HERO高校联盟知识峰会. 小白也能懂的迁移学习——算法和案例分析. 2020.09.24. [pdf]
• 南京智谷人工智能研究院. 人工智能技术及其应用. 2020.04.07. [pdf]

Awards & Honors

• 宝钢奖学金. 2024
• 江苏省第十八届大学生职业规划大赛一等奖·金奖. 2023
• 华为奖学金. 2023
• 南京大学学生年度人物提名奖. 2022
• 优秀研究生标兵. 2022
• 博士生国家奖学金. 2022
• 栋梁奖学金优秀奖（科研创新奖）. 2021
• 华为奖学金（华为-南京大学LAMDA人工智能联合实验室突出贡献奖）. 2021
• 优秀硕士研究生. 2020
• 硕士研究生英才一等奖学金. 2020
• IJCAI'19学生旅行奖（IJCAI'19 Student Travel Grant Reward）. 2019
• LAMDA学生会编辑部最佳原创三等奖. 2019
• 南京大学优秀毕业生. 2018
• 本科生国家奖学金. 2016
• 南京大学优秀学生. 2015

Projects

• 学件项目（华为2012、华为Model Arts）. 2022.03 - 2022.10
• 集成学习 & 分布式模型重用（华为2012）. 2021.06 - 2022.10
• 分布式模型复用技术--模型私有化（华为2012）. 2019.12 - 2020.12
• 基于联邦学习的多APP智能业务感知（华为NAIE）. 2019.11 - 2020.05
• 基于迁移学习和模型重用的通信质量增益预测（华为NAIE）. 2019.4 - 2019.9
• 基于特征工程和双层分类器的齿轮故障检测（远景能源）. 2019.3

Patents

• 一种模型训练方法及其相关设备. CN202210304574
• 一种基于联邦迁移学习的模型训练方法及计算节点. CN113159283B [已授权]
• 一种训练模型的方法、数据处理的方法以及装置. CN112651511B [已授权]
• 一种用于解决风机故障检测中标记噪音和标记不足的方法. CN112177865B [已授权]
• 一种基于自动示例选择的端到端多示例学习方法. CN112183752B [已授权]
• 一种基于零值比例频谱特征的机械传动故障检测方法. CN112347944B [已授权]
• 一种基于阶梯误差频谱特征的机械传动故障检测方法. CN112364762B [已授权]
• 通信领域不平衡业务感知的训练推理分歧消解方法和系统. CN202210172383
• 一种基于异构模型重用的跨区域通信流量预测方法. CN201910803496

Teaching Assistant

• 2021年秋季本科生/研究生高级机器学习课程. 詹德川.
• 2020年秋季本科生数字信号处理课程. 叶翰嘉.
• 2019年春季本科生数据结构课程. 姜远教授.

Correspondence

Email: lixc@lamda.nju.edu.cn
Office: Room 201, Yi-Fu Building, Xianlin Campus of Nanjing University
Address: National Key Laboratory for Novel Software Technology
                 Nanjing University, Xianlin Campus
                 163 Xianlin Avenue, Qixia District, Nanjing 210023, China